Impact of projected climate change on the hydrology in the

Impact of projected climate change on the hydrology in the
Hydrol. Process. 29, 4379–4397 (2015)
Published online 11 May 2015 in Wiley Online Library
( DOI: 10.1002/hyp.10497
Impact of projected climate change on the hydrology in the
headwaters of the Yellow River basin
Yueguan Zhang,1,2* Fengge Su,1 Zhenchun Hao,3 Chongyu Xu,4,5 Zhongbo Yu,2,6 Lu Wang7
and Kai Tong1
Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences,
Beijing 100101, China
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Hohai University, Nanjing 210098, China
College of Water Resources and environment, Hohai University, Nanjing 210098, China
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China
Department of Geosciences, University of Oslo, Oslo 0316, Norway
Department of Geoscience, University of Nevada, Las Vegas 89154, NV, USA
Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft 2600, The Netherlands
Located in the northeast of the Tibetan Plateau, the headwaters of the Yellow River basin (HYRB) are very vulnerable to climate
change. In this study, we used the Soil and Water Assessment Tool (SWAT) model to assess the impact of future climate change
on this region’s hydrological components for the near future period of 2013–2042 under three emission scenarios A1B, A2 and
B1. The uncertainty in this evaluation was considered by employing Bayesian model averaging approach on global climate
model (GCM) multimodel ensemble projections. First, we evaluated the capability of the SWAT model for streamflow
simulation in this basin. Second, the GCMs’ monthly ensemble projections were downscaled to daily climate data using the biascorrection and spatial-disaggregation method and then were utilized as input into the SWAT model. The results indicate the
following: (1) The SWAT model exhibits a good performance for both calibration and validation periods after adjusting
parameters in snowmelt module and establishing elevation bands in sub-basins. (2) The projected precipitation suggests a general
increase under all three scenarios, with a larger extent in both A1B and B1 and a slight variation for A2. With regard to
temperature, all scenarios show pronounced warming trends, of which A2 displays the largest amplitude. (3) In the terms of total
runoff from the whole basin, there is an increasing trend in the future streamflow at Tangnaihai gauge under A1B and B1, while
the A2 scenario is characterized by a declining trend. Spatially, A1B and B1 scenarios demonstrate increasing trends across most
of the region. Groundwater and surface runoffs indicate similar trends with total runoff, whereas all three scenarios exhibit an
increase in actual evapotranspiration. Generally, both A1B and B1 scenarios suggest a warmer and wetter tendency over the
HYRB in the forthcoming decades, while the case for A2 indicates a warmer and drier trend. Findings from this study can
provide beneficial reference to water resource and eco-environment management strategies for governmental policymakers.
Copyright © 2015 John Wiley & Sons, Ltd.
climate change; SWAT model; Bayesian model averaging; the headwaters of the Yellow River basin; Tibetan
Received 21 June 2014; Accepted 23 March 2015
Global climate change caused by the increasing concentration of carbon dioxide and other greenhouse gases
(GHGs) in the atmosphere received extensive attention in
recent years. The Fifth Assessment Report of the
International Panel on Climate Change states that global
air temperatures had increased by 0.85 °C over the period
*Correspondence to: Yueguan Zhang, Key Laboratory of Tibetan
Environment Changes and Land Surface Processes, Institute of Tibetan
Plateau Research, Chinese Academy of Sciences, Building 3, Courtyard
16, Lincui Road, Chaoyang District, Beijing 100101, China.
E-mail: [email protected]
Copyright © 2015 John Wiley & Sons, Ltd.
1880–2012 (IPCC, 2013). Global warming is likely to
have significant impacts on the hydrologic cycle and
affects water resources system (Chang and Jung, 2010;
Wang et al., 2011a, b; Jin and Sridhar, 2012; Ashraf
Vaghefi et al., 2014). Recently, the warming over the
Tibetan Plateau is generally more rapid than that of
surrounding areas (Zhou and Huang, 2012). With a mean
elevation of 4000 m, the headwaters of the Yellow River
basin (HYRB) is located in the northeastern Tibetan
Plateau. This region is called the ‘water tower’, as it
covers only 15% of the whole Yellow River basin but
contributes to 38% of the total runoff (Li et al., 2008). The
changing hydrologic regimes induced by climate change
and subsequently resulting variation in available water
resources will strongly influence not only the ecoenvironment and socio-economic development of the local
region but also the downstream areas. Therefore, climate
change impact studies in the headwaters of the Yellow
River have aroused widespread concern in the academic
community. For example, the historical observations on
precipitation, temperature and evapotranspiration have
been analysed to understand the trends of climate change
for the past few decades (Zhao et al., 2008; Xie et al.,
2010; Zhang et al., 2011; Yi et al., 2012). The general
conclusion is that temperature shows a significant warming
trend while the trend for precipitation is not obvious but a
drying direction is detected.
Meanwhile, in order to investigate future climate
change over the HYRB, most of the literatures involve
employing global climate models (GCMs) outputs to
drive hydrological models. For example, Xu et al. (2009)
used outputs from four GCMs and the Soil and Water
Assessment Tool (SWAT) model to investigate how
streamflow will be affected in the future; by using the
information from seven GCMs and a distributed
hydrological model, Li et al. (2008) analysed future
runoff characteristics in the HYRB; Zhang et al. (2012)
also applied the SWAT model to simulate the runoff and
examined the impact of climate change on this region,
utilizing Commonwealth Scientific and Industrial Research Organisation (CSIRO) and National Center for
Atmospheric Research (NCAR) climate models under
A1B and B1 scenarios. Also, Li et al. (2012) employed a
regional climate model to study future climate scenarios
and their effect on the water resources over the HYRB. In
addition, Wang et al. (2012) and Hu et al. (2013) used
predictors from GCMs to investigate the changes in
future precipitation and temperature extreme indices,
with the support of a statistical downscaling model.
However, most of the aforementioned works are based
on a limited number of GCMs, and there is little research
on ensemble projections of climate model outputs to
reduce the uncertainty in the climate change impact
studies. Meanwhile, in investigating hydrological response to future climate change, many researches only
deal with streamflow component, and less attention has
been paid to other hydrological variables such as
evapotranspiration and groundwater. Furthermore, little
effort has been made to systematically analyse the roles
of the snowmelt and establish elevation bands in the
hydrological simulation in this heterogeneous and
elevated mountainous range, although some hydrological
modelling has already been operated in this region (Xu
et al., 2009; Zhang et al., 2012; Cuo et al., 2013; Zhang
et al., 2013).
The forecast skill of multimodel ensemble mean is
superior to that of each ensemble member (Fritsch et al.,
Copyright © 2015 John Wiley & Sons, Ltd.
2000; Min et al., 2004; Palmer et al., 2004; Palmer
et al., 2005; Nohara et al., 2006; Phillips and Gleckler,
2006). The simplest way is the arithmetic ensemble
mean where each model is weighted equally. Reliability
ensemble averaging method developed by Giorgi and
Mearns (2002) is another method where the individual
GCM weights are derived from model performance and
future ensemble convergence. In recent years, Bayesian
model averaging (BMA) (Raftery et al., 2005) is
suggested and applied to GCM evaluation and multiGCMs ensemble, to consider model uncertainty systematically. BMA combines information from a series of
models to obtain a probability distribution for a quantity
of interest and accounts for model uncertainty. Owing to
its significant advantages, the BMA method has been
utilized by many researchers and found to be efficient in
reducing model biases and uncertainty for the GCM
projection (Min et al., 2007; Bhat et al., 2011; Yang
et al., 2011).
In this paper, we employed the SWAT model to
investigate the hydrological response to future climate
change in the headwaters of the Yellow River for the near
future period (2013–2042), based on the GCM ensemble
result by the BMA method. One main objective is to look
at the long-term mean changes in hydrological components including evapotranspiration, surface water and
groundwater over the focus area. The second objective is
to analyse the uncertainties related to GCM models and
emission scenarios in the projected hydrological variables. In addition, we also assess the roles of snowmelt
and elevation zones in hydrological modelling in the
HYRB, given that the focus area is located in the Tibetan
Plateau with high elevation and complex snowfall and
snowmelt processes.
Study area
The HYRB is located in 95.5–103.5°E and 32–36.5°N
with the Tangnaihai hydrological station as the control
outlet of the basin (Figure 1). It covers an area of about
121 973 km2 (15% of the whole Yellow River basin), and
the main river length is about 1553 km. The population
density is sparse, and so, this area can be regarded as
unimpaired with limited human activities. The average
elevation is about 4217 m a.m.s.l. and ranges between
2686 and 6137 m a.m.s.l. About 80% of this region is
covered by grassland, and the lakes and swamps account
for about 2000 km2. The HYRB belongs to the Tibetan
Plateau climate system, characterized by a wet and warm
summer and a cold and dry winter. From southeast to
northwest, annual average temperature varies between 4
and 2 °C.
Hydrol. Process. 29, 4379–4397 (2015)
Figure 1. Sketch map of the study area
Input data
Daily climate data from 1960 to 1999, including
precipitation, maximum, minimum and mean air temperature, solar radiation, wind speed and relative humidity,
were obtained from 23 meteorological stations in and
around the HYRB (Figure 1). The data quality for these
forcings has been controlled by the China Meteorological
Administration. In order to utilize the bias-correction and
spatial-disaggregation (BCSD) downscaling method of
Wood et al. (2002, 2004), all station data were
interpolated to the 1 × 1° resolution grids by using the
inverse distance weighting method. Because the SWAT
model requires gauge data, we treated each grid point as a
‘gauge’ according to the practice of Moradkhani et al.
(2010). The centroid of one grid cell is considered as the
location of a ‘gauge’.
In this study, GCMs in phase 3 of the Coupled Model
Intercomparison Project (CMIP3) under A1B, A2 and B1
scenarios were used for climate change projections. The
details of the models used are listed in Table I. Data
include monthly mean temperature and precipitation from
1960 to 2042. These models have different spatial
resolutions in the range of about 1–4°. To obtain the
ensemble results of different models and facilitate the
employment of the BCSD method, these GCMs data were
interpolated onto a common 2 × 2° grid.
Besides climate data, SWAT also requires a great deal
of spatial data such as digital elevation model, land
use/cover and soil data. Detail information for these data
Copyright © 2015 John Wiley & Sons, Ltd.
is presented in Table II. Daily observed river discharges at
the Tangnaihai control station were used to evaluate the
SWAT model simulations. In this study, the historical
period 1961–1990 is defined as the baseline period,
because it incorporates some of the natural variability of
the climate, including both dry (1970s) and wet (1980s)
times (Prudhomme et al., 2002).
SWAT model
SWAT is a computationally efficient simulator of
hydrology and water quality at various scales. The model
includes procedures to describe how carbon dioxide
concentration, precipitation, temperature and humidity
affect plant growth, evapotranspiration, snow and runoff
generation among other variables and, therefore, is
usually used to study climate change impacts (Arnold
et al., 2007). SWAT has been successfully applied to
many parts of the world and proved to adequately
reproduce hydrological processes of watersheds across a
range of geographical regions and climates (Jha et al.,
2004; Borah et al., 2006; Setegn et al., 2011; Zhang et al.,
Snowmelt processes in SWAT model
SWAT uses a temperature index-based approach to
estimate snowmelt processes. Snowmelt is controlled by
Hydrol. Process. 29, 4379–4397 (2015)
Table I. List of global climate model simulations
SRES A1B(18)
MIROC3.2 medres
SRES A2(16)
MIROC3.2 medres
MIROC3.2 medres
the air and snow pack temperature, the melting rate and
the area coverage of snow. The model considers melted
snow as rainfall in order to compute runoff and
percolation. Snowmelt is estimated as a linear function
of the difference between the average snow pack
maximum air temperature and the user-defined snowmelt
temperature threshold (Neitsch et al., 2005)
SNOmlt ¼ bmlt snocov T snow þ T mx
T mlt
where SNOmlt is the amount of snowmelt on a given day
(mm), bmlt is the melt factor for the day (mm H2O/day/°C),
snocov is the fraction of hydrologic response unit area
covered by snow, Tsnow is the snow pack temperature on
a given day (°C), Tmx is the maximum air temperature on
a given day (°C) and Tmlt is the threshold temperature
above which snowmelt is allowed (°C). The melt factor is
allowed seasonal variation with maximum and minimum
Germany, Korea
4.0 × 3.0L12
5.0 × 4.0L20
5.0 × 4.0L20
5.0 × 4.0L21
2.5 × 3.75L19
2.5 × 3.75L19
values occurring on summer and winter solstices (Fontaine
et al., 2002):
bmlt6 þ bmlt12
bmlt6 bmlt12
bmlt ¼
ðd n 81Þ
where bmlt is the melt factor for the day (mm H2O/day/°C);
bmlt6 and bmlt12 are the melt factors for 21 June and 21
December (mm H2O/day/°C), respectively; dn is the day
number of the year.
Elevation bands algorithm in SWAT model
Elevation is an important factor in dictating the
variation of temperature and precipitation (Zhang et al.,
2008). To account for orographic effects on climatic
variables, SWAT allows up to 10 elevation bands to be
split in each sub-basin. The addition of elevation bands
Table II. Data used and sources in Soil and Water Assessment Tool (SWAT) model
Digital elevation National Geomatics Center of China
model (DEM)
Soil data
Institute of Soil Science, Chinese
Academy of Sciences (CAS)
Land use data
Cold and Arid Regions Environmental
and Engineering Research Institute, CAS
Weather data
China meteorological data sharing
service system
Flow data
Water Resources Conservancy Committee
of the Yellow River basin
Copyright © 2015 John Wiley & Sons, Ltd.
1 : 250 000
3 arc sec
1 : 1,00 000 1 km
1 : 100 000
1 km
Classified soil and physical properties
such as bulk density and texture
Classified land use such as cropland
and pasture
Precipitation, air temperature, wind
speed, relative humidity and solar radiation
River flow at Tangnaihai gauge
Hydrol. Process. 29, 4379–4397 (2015)
allows SWAT to better represent the distribution of
precipitation and temperature over areas that contain large
elevation range. Precipitation and temperature are estimated for each band as a function of the respective lapse
rate and the difference between the gauge elevation and
the average elevation specified for the band (Fontaine
et al., 2002). The temperature and precipitation for each
band were adjusted using the following two equations:
T B ¼ T þ ðZ B Z ÞT laps
PB ¼ P þ ðZ B Z ÞPlaps
where TB is the elevation band mean temperature (°C), T
is the temperature measured at the weather station (°C),
ZB is the midpoint elevation of the band (m), Z is the
weather station’s elevation, PB is the precipitation falling
in the elevation band (mm H2O), P is the precipitation
measured at the weather station, and Tlaps and Plaps are the
temperature lapse rate (°C/km) and precipitation lapse
rate (mm/km), respectively.
Bayesian model averaging method
We employ BMA to derive future climate projections.
BMA has recently been proposed as a way of correcting
under dispersion in ensemble forecasts (Raftery et al.,
2005; Min and Hense, 2006). Rather than choosing a
single model among the set of models to use for
prediction, the prediction of a variable y- (precipitation
or temperature) is instead conditioned on the entire set of
pðyÞ ¼ ∑ p yM k Þp M k yT Þ
where p(y|Mk) is the forecast probability density function
based on model Mk (GCMs), estimated from the training
data; p(Mk|yT) is the posterior probability of model Mk
being corrected given the training data.
Through a series of transformation, the BMA predictive mean can be derived:
E yf 1 ……f k ¼ ∑ wk ðak þ bk f k Þ
This can be viewed as a deterministic forecast and
expected to be more skilful than either the ensemble mean
or any one member. The BMA weights and standard
deviation are estimated by maximum likelihood method.
Generally, the longer the training period, the better the
BMA parameters are estimated (Raftery et al., 2005).
In this study, the observed data sets of surface air
temperature and precipitation for the years 1960–1999
have been used to train BMA weights for GCM models
Copyright © 2015 John Wiley & Sons, Ltd.
under 20C3M emission scenario. Next, on the basis of
these weights and parameters, future scenarios of climatic
variables were generated for the period of the years 2013–
2042 under scenarios A1B, A2 and B1 in the HYRB. So
far, the time scale of future GCM forcings is still in
month, with a spatial resolution of 2° × 2°. However, daily
meteorological data are required as input to force the
SWAT model for hydrological simulation. Downscaling
method is necessary to link the ensemble GCM output at
large-scale to small-scale climate forcing used for the
hydrological model, which will be detailed in the
following part.
Downscaling method
In this study, the BCSD approach was utilized to
downscale the GCM results. As a statistical downscaling
method, BCSD has been widely used and tested in the
climate change analysis (Maurer et al., 2009; Bennett
et al., 2012). The BCSD method has been shown to
provide downscaling capabilities comparable with those
of other statistical and dynamical methods in the context
of hydrologic impacts (Wood et al., 2004). From the
studies of Wood et al. (2004), when applied to the
National Center for Atmospheric Research/Department of
Energy (NCAR/DOE) Parallel Climate Model and
Regional Climate Model outputs, the BCSD method can
successfully reproduce the main features of the observed
hydrometeorology from the retrospective climate simulation, whereas linear interpolation method leads to
unacceptably biased hydrologic simulation. In this study,
we employed the BCSD method to downscale the BMAbased monthly GCM forcings of 2° × 2° to future climate
data with 1° × 1° resolution, which consist of daily mean
temperature and daily precipitation for 2013–2042 under
scenarios A1B, A2 and B1.
Based on observed data for 1960–1999, we have
established the linear relationships between daily mean
temperature and daily maximum–minimum temperature
at each grid. According to statistical analysis, the
determination coefficient R2 reached more than 0.9 for
both relationships. Then the future daily maximum and
minimum temperature were obtained from such relationships and BCSD-based daily mean temperature. Meanwhile, SWAT includes the WXGEN weather generator
model (Sharpley and Williams, 1990) to generate climatic
data or fill in gaps in measured records (Neitsch et al.,
2005). Therefore, solar radiation, relative humidity and
wind speed can be generated for the future scenarios
using this built-in weather generator.
Model performance evaluation
In this study, we followed Nash–Sutcliffe efficiency
(NSE), root-mean-square error observations standard
Hydrol. Process. 29, 4379–4397 (2015)
deviation ratio (RSR) and percent bias (PBIAS) (Moriasi
et al., 2007) as SWAT evaluation statistics. SWAT
performance can be judged (Table III) according to the
work of Moriasi et al. (2007).
n sim 2
∑ Y obs
i Yi
NSE ¼ 1 i¼1
n mean 2
∑ Y obs
n 2
Y sim
∑ Y obs
RSR ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n mean 2
∑ Y obs
i Y
n obs
∑ Y sim
i Yi
PBIAS ¼ i¼1
Y obs
where Y obs
and Y sim
are the observed data and simulated
value at time i; Ymean is the mean of the observed data for
the whole evaluating period.
The effects of snowmelt processes and elevation bands on
the simulation
As belonging to the Tibetan Plateau climate system, the
HYRB is characterized by a long cold winter and a short
warm summer. The snow accumulation process generally
starts in October, lasting to March of the following year,
and then from April to June, the snow pack begins to
release meltwater. In addition, high elevation, low
temperature and a large area make the snow accumulation
and melt processes significant. Also, elevation is an
important factor dictating the variation of many meteorological variables, such as temperature and precipitation.
Therefore, in this study, one modelling objective involves
considering the effects of snowmelt processes and
elevation bands on the SWAT performance in the HYRB.
Also, some effort has been made to adjust soil and
groundwater parameters to further improve modelling
accuracy in this region.
Figure 2 shows the long-term monthly observed and
simulated flow at the Tangnaihai gauge at different phases
of simulation for 1961–1990. The major problems
identified in the initial setup were that the rising
hydrograph limb began too late and there existed a
systematic underestimation of the observed flow. Based
on the classification of model efficiency in Table III, the
NSE of 0.51 (Table VI) for the initial simulation indicated
an unsatisfactory result.
Table IV lists the parameters used in the snowmelt
routines, and simultaneously, some parameters are given
simple descriptions as follows. SNO100 is the threshold
depth of snow and depends on factors such as vegetation
distribution and wind loading of snow. The melt factor for
snow on 21 June is parameterized by bmlt6, which
represents the maximum melt rate. The increase of bmlt6
can accelerate melting of the snow pack. As there are little
available data from meteorological stations to be used for
setting parameters in the snowmelt routines, parameters
for this algorithm were estimated based on the studies of
Fontaine et al. (2002) and Zhang et al. (2008). Figure 2
shows the effect of adjusting snow parameters on the
hydrologic modelling in the HYRB. Compared with the
initial result, the simulated flow is increasing and closer to
the observed value. As a result, the PBIAS was improved
to 16%, although there is still a considerable bias. The
NSE of 0.75 indicated a better simulation in this focus
area. According to Table III, the modelling of SWAT
achieved a satisfactory result.
As the HYRB has a large elevation variation, adding
elevation bands could better represent the spatial
variability of climate. In this study, five elevation bands
were established within each sub-basin. The average
elevation of each band and the fraction of sub-basin area
within that band were also to be specified. The
temperature lapse rate TLAPS adopted a commonly used
value of 6.5 °C/km. However, it was difficult to estimate
the precipitation lapse rate PLAPS because of the low
density of the climate data in the HYRB. Thus, this
parameter in this analysis was chosen based on the
Table III. Statistics for simulation result performance ratings*
Performance rating
Very good
0.00 ≦ RSR ≦ 0.50
0.50 < RSR ≦ 0.60
0.60 < RSR ≦ 0.70
RSR > 0.70
0.75 < NSE ≦ 1.00
0.65 < NSE ≦ 0.75
0.50 < NSE ≦ 0.65
NSE ≦ 0.50
PBIAS < ±10
±10 < PBIAS < ±15
±15 ≦ PBIAS ≦ ±25
PBIAS ≧ ±25
NSE, Nash–Sutcliffe efficiency; RSR, root-mean-square error observations standard deviation ratio; PBIAS, percent bias.
*From Moriasi et al. (2007).
Copyright © 2015 John Wiley & Sons, Ltd.
Hydrol. Process. 29, 4379–4397 (2015)
Figure 2. Long-term monthly simulated flow versus observed flow for different stages corresponding to the effects of snow parameters, adding elevation
bands and adjusting other important parameters
findings of Fontaine et al. (2002) and was finally set at
+0.5 mm/km. After using the elevation band algorithm, it
can be identified from Figure 2 that the systematic
underestimation problem in the early simulations was
improved, with a PBIAS of +8%. The NSE of 0.77
reflected a good fit between the modelled and observed
hydrograph. From the classification criteria listed in
Table III, this simulation was within the ‘good’
performance rating.
Calibrating soil and groundwater parameter
After considering the roles of snowmelt processes and
elevation bands in the simulation, it is necessary to adjust
soil and groundwater parameters because of the significant change in water recharge related to snowmelt and
orographic effects. In this study, monthly discharge
records of Tangnaihai station for the years 1961–1990
were split into two segments with 1961–1980 for
calibration and 1981–1990 for validation. Manual
calibration process, i.e., the trial and error method, was
employed to improve model accuracy.
These soil and groundwater parameters mainly
consisted of soil evaporation compensation factor
(ESCO), base-flow recession constant (ALPHA_BF),
threshold water level in shallow aquifer for evaporation
(REVAPMN) and delay time for aquifer recharge
(GW_DELAY). By using trial and error techniques and
based on studies such as Zhang et al. (2008) and Neitsch
et al. (2005), the parameter values were determined
(Table V). Figure 2 shows the effect of adjusting
groundwater parameters on the hydrologic modelling in
the HYRB, which indicates a good simulated result for
the whole period. In addition, Figure 3 also shows the
SWAT performance in simulated monthly streamflow for
the calibration period of 1961–1980 and validation period
of 1981–1990. Meanwhile, Table VI lists the evaluation
statistics, with NSE of 0.91 and 0.89 for calibration and
Table IV. Parameter descriptions and the values adopted in the snowmelt module
Adopted value
Snowfall temperature (°C)
Threshold depth of snow at 100% coverage (mm H2O)
Fraction of snow volume represented by SNO100 that corresponds to 50% snow cover
Snowmelt base temperature (°C)
Maximum snowmelt factor for 21 June (mm H2O⁄°C⁄day)
Minimum snowmelt factor for 21 December (mm H2O⁄°C⁄day)
5 to 5
5 to 5
Table V. Some soil and groundwater parameters for the final simulation based on manual calibration
Default value
Adopted value
Soil evaporation compensation factor
Base-flow recession constant
Delay time for aquifer recharge (days)
Threshold water level in shallow aquifer for evaporation (mm)
Copyright © 2015 John Wiley & Sons, Ltd.
Hydrol. Process. 29, 4379–4397 (2015)
Figure 3. Model results (monthly outputs) at the Tangnaihai gauge in calibration and validation
Table VI. Model performance evaluation for the initial setup,
calibration and validation period
Initial setup
Calibration (1961–1980)
NSE, Nash–Sutcliffe efficiency; RSR, root-mean-square error observations standard deviation ratio; PBIAS, percent bias.
validation periods, respectively. It can be seen that for the
focus area, model performances during calibration and
validation are very good after altering snow parameters,
using elevation bands and adjusting groundwater
As located in the northeastern Tibetan Plateau, the
HYRB exhibits a series of typical characters in the alpine
area, such as extreme elevation gradients, low temperature, and important snowfall and snowmelt processes.
Large-scale hydrologic modelling in mountainous terrain
is difficult because of irregular topography and poor data
resolution. In addition, rates of change in precipitation
and temperature with respect to elevation also complicate
the simulation of alpine hydrology. Hence, information
gained from this study may be used as a reference for
similar regions with complex terrain, in order to obtain a
better modelling result with the SWAT model.
GCM projections for the HYRB
Before we do projections for future climate change, it is
necessary to check the correlation between observed
climate data and ensemble GCMs model output based on
BMA in the baseline period. Figure 4 shows the scatter
plots for the comparison of the observed and GCMs
simulated monthly precipitation and temperature in the
baseline period of 1961–1990. The values of determination coefficient (R2) are 0.84 and 0.98 for monthly
precipitation and temperature, respectively. This suggests
a good agreement between ensemble GCMs output and
observed climate data.
Based on the BMA method, multimodel ensemble
projected future changes for temperature and precipitation
over the HYRB are presented in this section. Temperature
changes were given in degrees Celsius, and precipitation
changes were given as a percentage change, relative to
that of the baseline period (1961–1990). In the aspect of
spatial scale, changes for these factors were quantified at
Figure 4. Scatter plots for historic observed and ensemble global climate models modelled climatology in baseline period: (a) precipitation and (b)
Copyright © 2015 John Wiley & Sons, Ltd.
Hydrol. Process. 29, 4379–4397 (2015)
the sub-basin. From the temporal perspective, the relative
changes of monthly climatic variables were investigated
from January to December.
Precipitation changes
Figure 5 shows the historical precipitation distribution
and the anomaly maps for the HYRB under all the three
scenarios (A1B, A2 and B1). While all scenarios show an
increase in the northern parts, there are major differences
in the southeastern parts where the A1B and B1 scenarios
indicate an upward trend, but a downward tendency is
observed for A2. Under the impact of the Asian summer
monsoon, the annual precipitation in the southeastern part
of the HYRB is over 500 mm, and thus, this zone is the
wet area of the focus region. Therefore, precipitation
variation in this part has a crucial effect on the whole
water amount of the HYRB. This phenomenon can be
further corroborated especially under A2 scenarios. In this
case, although a modest to large positive precipitation
anomaly (3–11%) appears in most of the HYRB, the
small negative anomaly (around 1.8%) in the wet part
finally leads to a slight increase of only 2% over the
whole HYRB at the end of the year 2042. At the same
time, the other two situations (A1B and B1) display a
positive anomaly in the southeastern area and finally
result in larger increments, with 14.4% (A1B) and 9.7%
(B1) over the HYRB. In the water-limited places such as
the extreme northwestern region, one part of the driest
areas, both A1B and A2 scenarios show the largest
increases in precipitation with magnitude of 19% and
11%, about 74 and 45 mm in absolute annual amount,
respectively. But B1 exhibits the smallest increase of just
about 2% (about 7 mm) over this part. Projected
Figure 5. The anomaly map of precipitation averages. (a) The historic precipitation distribution. (b) Results of scenario A1B. (c) Results of scenario A2.
(d) Results of scenario B1
Copyright © 2015 John Wiley & Sons, Ltd.
Hydrol. Process. 29, 4379–4397 (2015)
increasing precipitation in this significantly water-scarce
region may alleviate the local continuous droughts to
some extent and has a positive effect on the local ecology
Figure 6a depicts the relative change (%) of predicted
long-term average precipitation to the historical data for
different scenarios. Under A1B, an increasing trend can be
found in all months ranging from the minimum quantity of
5% in June to the maximum value of 47% in October. For
A2, decreases can be found in April, August, September
and November, while an upward trend occurs in the
remaining months. A similar pattern of change with A1B
can be found for B1. In terms of seasonal scale, the largest
increase will occur in winter under A2 and B1 scenarios,
while for A1B, the fall season will experience the largest
growth and then followed by winter.
Surface air temperature changes
Figure 7 shows BMA ensemble GCMs projected
changes in annual temperature between the baseline
(1961–1990) and future (2013–2042) periods. A general
increasing temperature can be found over the HYRB in
the future. Compared with precipitation, spatial patterns
of trends in temperature have a much higher degree of
consistency. All the three scenarios display a small
increase in the relatively warm southeastern part and a
more pronounced increase in the very cold northwestern
area of the HYRB. According to Figure 7c, under the A2
scenario, the ordinary warming amplitude can reach about
1 °C, and the largest warming is expected within the
southern part with a magnitude of up to 1.3 °C.
The average monthly changes in surface air temperature
are presented in Figure 6b. Consistent increases from January
to December can be observed under the three scenarios.
Almost all months will experience around 1 °C warming for
both A1B and A2 scenarios in the future. According to the
statistical results, the increment for temperature is around
1.15 °C (A1B), 1.27 °C (A2) and 1.08 °C (B1) in the end of
2042. This outcome is consistent with the ranking of the
changes for the GHG level: The smallest increase is for the
lowest-emission B1 scenario, and the largest increase is for
the highest-emission A2 scenario, which also suggests that
the warming magnitude is sensitive to the emission scenarios
Figure 6. Comparison of changes for different elements between historic and future climate conditions: (a) precipitation, (b) temperature, (c) actual
evapotranspiration (AET), and (d) total runoff
Copyright © 2015 John Wiley & Sons, Ltd.
Hydrol. Process. 29, 4379–4397 (2015)
Figure 7. Changes of temperature over the HYRB for the period of 2013–2042. (a) The historic absolute values. (b) Results of scenario A1B. (c) Results
of scenario A2. (d) Results of scenario B1
in the HYRB. In the seasonal scale, the increase in winter is
the largest among the four seasons under all scenarios based
on statistical analysis.
Impact of climate change on actual evapotranspiration
Besides precipitation and temperature, evapotranspiration
is another very important climatic factor that controls the
energy and mass exchange between terrestrial and atmosphere and plays a key role in hydrologic processes.
Evapotranspiration exceeds runoff in most river basins and
on all continents except Antarctica (Dingman, 1994). Actual
evapotranspiration (AET), comprising the actual evaporation
(the non-productive part) and the actual transpiration (the
productive part), is commonly referred as the ‘green water’
(Abbaspour et al., 2009). In Figure 8, the average values of
AET based on the historic data and the anomaly graphs for
the three scenarios are shown for the HYRB. The differences
Copyright © 2015 John Wiley & Sons, Ltd.
are calculated between the averages of the 2013–2042 period
and those of the 1961–1990 period. Generally, all scenarios
predict an increase in actual evapotranspiration across the
HYRB. The future simulations show a small increase in the
extreme southeastern and northwestern parts, while the semiarid area, such as the central and northern parts, will
experience a larger increase. However, it should be noted that
an increase of >20% in the north of the focus region amounts
to about 20 mm/year, which is still very small in respect of
quantity but could have a substantial impact on the local
ecosystem of the water-scarce area.
Figure 6c exhibits the results for the relative changes of
monthly mean AET under A1B, A2 and B1 scenarios. It is
noted that monthly mean AET consistently has an increasing
trend among the 12 months under all scenarios. In particular,
the largest increase of about 30% can be found in October
under A2, while a slight increase of less than 6% will occur in
July and August under the three scenarios. In addition, spring
Hydrol. Process. 29, 4379–4397 (2015)
Figure 8. Relative changes of actual evapotranspiration (AET) between historic and future climate conditions. (a) The historic absolute values. (b)
Results of scenario A1B. (c) Results of scenario A2. (d) Results of scenario B1
months (March–May) exhibit a relatively large increase of
between 15% and 26% for all the three scenarios.
Evapotranspiration (ET) has an important impact on the
available water resources in this semi-arid and semi-humid
region. The trends of increase in ET are probably attributed to
the increase in the air temperature. In this study, the Penman–
Monteith equation was employed to calculate evapotranspiration that can be affected greatly by the minimum and
maximum temperatures. This is consistent with the findings
of Setegn et al. (2011), which indicated that the variation in
AET is pertinent to the changes in temperature.
Impact of climate change on surface runoff and
The anomalies relative to the baseline period for surface
runoff and groundwater over the HYRB are listed in
Figures 9 and 10, respectively. It can be seen that, for A1B
Copyright © 2015 John Wiley & Sons, Ltd.
scenario, the surface runoff generally shows an increasing
trend, and most of the area will enjoy more than 20%
growth, especially part of the northwestern zone showing a
magnitude of even 40% or larger. B1 indicates a similar
situation with A1B and differs only in the increasing
amplitude. In contrast, the A2 scenario suggests a decline
in the surface runoff across the entire region, and most subbasins will suffer a large decrease of more than 20%.
Groundwater is an important runoff component for the
HYRB, which can account for 65% of the total runoff
(Chen et al., 2008). Generally, it is observed from
Figure 10 that groundwater flow increased under A1B
and B1 but reduced for A2 in the future. With regard to
changes in space, both A1B and B1 indicate similarly
more increments in the wet southeastern part of the study
region, whereas a different result in the extreme
northwestern area can be detected, which indicates a
large increase in A1B scenario but a significant descent
Hydrol. Process. 29, 4379–4397 (2015)
Figure 9. Relative changes of surface runoff between historic and future climate conditions. (a) The historic absolute value. (b) Results of scenario A1B.
(c) Results of scenario A2. (d) Results of scenario B1
for B1 case. This is alike the spatial changes in the
projected precipitation in the previous section. At the same
time, the simulation for A2 implies that almost the entire
region except for a small northwestern part will
experience a moderate to substantial decrease (more than
34%) in groundwater. Because it occupied a larger
proportion of the total runoff, the fluctuation in
groundwater could impose a significant impact on the
total runoff, as discussed in the following section.
Impact of climate change on the total runoff
Figure 11 exhibits the historic total runoff distribution
and the anomaly maps over the HYRB under three
scenarios. The spatial distribution of total runoff has a
similar pattern with that in the groundwater. Generally,
A1B and B1 scenarios demonstrate increasing trends
across most of the region. But under scenario A2, a large
Copyright © 2015 John Wiley & Sons, Ltd.
decrease ranging from about 33% to 11% can be
found in the majority of the HYRB, and the prominent
generated-runoff area, i.e., the southeastern part from
Jimai to Maqu, will also experience a considerable
decrease of about 15%. Such a large decrease in total
runoff is probably caused by changes in actual evapotranspiration, as a result of rapid warming temperature.
The relative changes in monthly total runoff are shown
in Figure 6d. Under A1B, it is noticed that all the months
exhibit increasing trends, and the magnitude varies from
5.3% to 40%. Seasonally, there are large increases in spring
and autumn, which suggests more floods in the future. B1
scenario displays a similar pattern with the A1B except for
decreasing total runoff in June and July. On the contrary,
total runoff under A2 is reduced through all the months,
and the maximum decrease of 16% occurs in April to June.
This indicates that more serious and frequent droughts will
occur in late spring and early summer in the near future.
Hydrol. Process. 29, 4379–4397 (2015)
Figure 10. Relative changes of groundwater between historic and future climate conditions. (a) The historic absolute values. (b) Results of scenario A1B.
(c) Results of scenario A2. (d) Results of scenario B1
In summary, this study revealed a hydrological response
to the near future climate change in the HYRB. In this
particular region, the total runoff is more vulnerable than
other hydrological variables with respect to the enhanced
GHG loading, because it is the key variable for the
mountain ecology and sustainable development of the
regional alpine steppe and alpine meadows. The HYRB is
also vulnerable to the increase of temperature caused by
enhanced GHG loading, which will intensify the degradation process of the permafrost and lead to significant
alternations in regional ecosystem and water cycling.
Implication of climate change for the regional runoff
In this study, the impact of climate change on future
precipitation, temperature and some hydrological eleCopyright © 2015 John Wiley & Sons, Ltd.
ments has been investigated using CMIP3-CGCM
multimodel ensemble projections (2013–2042). According to statistical analysis results, there is an increasing
trend in the future streamflow at Tangnaihai gauge under
A1B and B1 scenarios, with magnitude of 19% and 9.3%
relative to baseline period, respectively. There are large
increases in future precipitation for both A1B and B1
scenarios, which is the prominent reason for the raising
runoff in the HYRB. Results from other studies also
indicate that the runoff will increase in the future under
A1B scenarios in the upstream of the Yellow River and
Brahmaputra River (Immerzeel et al., 2010; Li et al.,
2013). So this study supplements and further supports
their investigations by using the BMA method and the
SWAT model. However, there is a large decreasing trend
of 12.9% in discharge under scenario A2, even if some
increases are observed in future precipitation. One
possible explanation is that the slightly increased
Hydrol. Process. 29, 4379–4397 (2015)
Figure 11. Relative changes of total runoff between historic and future climate conditions. (a) The historic absolute values. (b) Results of scenario A1B.
(c) Results of scenario A2. (d) Results of scenario B1
precipitation has been compensated by more increase of
evapotranspiration, as a result of the rapid warming over
the HYRB. On the other hand, precipitation spatial
distribution for A2 may also have impact on the total
runoff. As indicated in Figure 5c, a moderate positive
precipitation anomaly appeared in the north and west part
(the dry area), while the southeastern region (wet area)
exhibited a negative anomaly. As energy in the dry zone
is sufficient for forcing evaporation, therefore, increased
precipitation is mostly dissipated by evaporation and thus
contributes to a litter change of the runoff over this
region. Consisting of sub-basin from Jimai to Maqu, the
wet region contributes to 55.7% of the total runoff over
the HYRB, although it only accounts for 33.7% of the
entire area. So the runoff change in this area plays a
crucial role in the total discharge of the HYRB. In the wet
region, the decreasing precipitation in the A2 scenario
along with the increased evaporation due to warming
Copyright © 2015 John Wiley & Sons, Ltd.
temperature causes the decreased runoff over this part. As
a result, the little contribution from the dry region and the
decreased water yield in the wet part finally lead to a
decline in runoff under scenario A2, despite the
increasing precipitation over the HYRB.
Meanwhile, different CMIP3 scenarios represented
different climate and hydrology projections in distinct
regions and hence indicated varied implications for these
areas. Compared with the small increase of runoff in the
water-limited northwestern part, the large decrease of
runoff in the wet southeastern area is considered more
significant for the focus region. On the one hand, the large
decrease of runoff will adversely impact the local
ecosystem in the southeast of the HYRB. The decreasing
runoff and an increase in the aridity index will degrade
the regional alpine steppe and alpine meadows. On the
other hand, as discussed earlier, the runoff changes in this
wet region play a very important role in the total
Hydrol. Process. 29, 4379–4397 (2015)
discharge of the HYRB. Therefore, the large decreasing
runoff in the southeast would not only affect the local
water resources but also reduce the total discharge from
the HYRB (‘water tower’ of the whole Yellow River
basin), which would inevitably have a crucial adverse
influence on the economy and livelihoods of people in the
downstream area of the Yellow River basin.
Implications of climate change for the local ecosystems
In recent years, the alpine grasslands in the HYRB
have suffered from severe degradation. Studies in the
source regions of the Yellow River showed that middle
and high-cover high-cold steppe areas decreased by
23.65%, and high-cover high-cold meadow areas
reduced by 6.85% from 1985 to 2000 (Qian et al.,
2006). Climate change is considered as an important
factor in the degradation of the grassland in the HYRB
(Gao et al., 2010; Wang et al., 2011a; Gao et al., 2014;
Yin et al., 2014). According to the analysis results of
meteorological data for 1961–1999 over the HYRB, the
study area has displayed a clear warming trend with an
annual air temperature rising of 0.2 °C per 10 years.
Meanwhile, the summer precipitation showed a decreasing tendency during this period. The limited precipitation and an increase in the aridity index degraded the
alpine steppe and alpine meadows among 1961–2000
(Zhou et al., 2005). Simultaneously, such warming and
drying climate affected the normal growth and reproduction of grass in the vigorous period, which further
exacerbated the grassland degradation. In this study,
both A1B and B1 scenarios suggested a warmer and
wetter climate over the HYRB in the near future. The
large increases of precipitation will benefit the recovery
of alpine grasslands and slow the degrading process to
some extent. In addition, the rising rainfall in summer as
represented in Figure 6a (about 34 mm for A1B and
21 mm for B1) will be advantageous to the growth of
grassland in the vigorous period, because water is one of
the important limiting factors to plants in the HYRB.
Hence, the local alpine ecosystems will also profit from
the more grass productivity during the growth period
and as a result may ease the degrading trend. However,
the warmer and drier climate under A2 scenario would
adversely impact the grassland ecosystem, and some
adaptive measures should be made to alleviate the
negative impacts.
Meanwhile, there are permafrost-related problems in
the headwaters of the Yellow River, especially in the
northwest part. The former studies on environmental
changes in this region have indicated that climatic
warming has led to deepen the active layer and resulted
to severe degradation of the permafrost region (Wu
et al., 2000; Zhang et al., 2010). Frozen soil layer, on
Copyright © 2015 John Wiley & Sons, Ltd.
the one hand, can serve as an impermeable layer and
obstruct soil liquid water infiltration and thus increase
soil water in the rooting zone; on the other hand, it can
concentrate various nutrients percolated from the above
active layer and thereby improve the nutrient supply
capacity of the soil. So the degradation of the permafrost
will inevitably lead to significant alternations in the
regional ecosystem and water cycling (Wang et al.,
2009; Wang et al., 2011a). Increasing permafrost
degradation decreases the number of plant families and
species, with hygrophytes and mesophytes gradually
replaced by mesoxerophytes and xerophytes in the
Qinghai–Tibet Plateau (Yang et al., 2013). In addition,
the thickness of the active layer in the permafrost has
an inverse correlation with the vegetation cover of the
alpine cold meadow and the alpine cold swamp
meadow (Wang et al., 2006; Wang et al., 2011a).
Therefore, the increasing thickness of active layer
induced by warming temperature will lead to reduction
of grassland area. This study shows that projected
temperature under all three scenarios exhibits prevailing
warming magnitude of at least 1 °C in the following
decades, so the rapid warming in the near future would
accelerate the degradation of the permafrost and
consequently would have a strong effect on the
ecosystems in this region.
Sources of uncertainty
There is a cascade of uncertainty in the hydrological
impact study of climate change, including GCMs, GHG
emissions scenarios, downscaling methods, hydrological
model structure and parameters (Wilby and Harris, 2006).
In our study, only two major sources of uncertainty from
GCMs and emission scenarios were investigated, and
other uncertainties were omitted. Recently, a number of
studies have investigated systematically the multiple
sources of uncertainty related to hydrological modelling
and changing climate (Boé et al., 2009; Forbes et al.,
2011; Majone et al., 2012; Bosshard et al., 2013).
Generally speaking, GCMs were the dominant source of
uncertainty for hydrological impacts (Wilby and Harris,
2006; Kay et al., 2009; Paton et al., 2013). However,
other uncertainty components such as downscaling
methods and GCM initial conditions were also found to
have similar or even larger importance to uncertainty
envelope for some hydrological variables (Horton et al.,
2006; Chen et al., 2011; Teutschbein et al., 2011). Also,
the investigation from Bastola et al. (2011) indicated that
the role of hydrological model uncertainty in climate
change studies is remarkably high and should be routinely
considered in impact studies. Therefore, a complete
analysis of uncertainty in climate impact studies is a next
important work in our study.
Hydrol. Process. 29, 4379–4397 (2015)
In this study, we assessed the impacts of projected
climatic change on the hydrological processes and water
resources variability in the HYRB by using the SWAT
model, the BCSD downscaling method and the state-ofthe-art BMA approach.
First, the role of adjusting snow parameters and setting
elevation bands has been evaluated in improving the
performance of the SWAT model in the HYRB. The
outcome is that the application of the temperature indexbased approach could seemingly lead to a satisfactory
hydrologic simulation, provided that the snow parameters
are well adjusted. Meanwhile, the application of temperature index plus elevation band gives a better simulation
By using the results from an ensemble of 16 or more
CMIP3-CGCMs, we investigated the projection of future
precipitation and temperature over the HYRB. Generally,
annual mean precipitation is projected to increase for all
three emission scenarios, with a large increment for A1B
(14.4%) and B1 (9.7%) but a slight change for A2 (2%).
In terms of air temperature change, the HYRB would
experience a warming trend in the future under the three
scenarios, with the most increases in A2 (1.27 °C) and
followed by A1B (1.15 °C) and B1 (1.08 °C).
With the employment of the SWAT model, we
investigated how changes in temperature and precipitation
might propagate into changes in total runoff and other
hydrological elements. Under A1B and B1 scenarios, the
total runoff for the near future indicated increasing trends
with magnitude of 19% and 9.3%. However, the total
runoff decreased by 12.9% for A2 scenario. In addition,
the changes in evapotranspiration, surface runoff and
groundwater were also examined, and it was found that
changes in groundwater flow might play a crucial role in
runoff variation because it contributes to over 60% of the
total runoff.
Although preliminary uncertainty analysis in this paper
was investigated including GCMs and emissions scenarios, other sources of uncertainty such as downscaling
methods, hydrological parameters and land use or land
cover change should also be given more attention. Thus,
more researches in our continuing work, especially a
thorough investigation of impacts from different uncertainty on the climate change impact analysis, are indeed
needed for a better understanding of the future changes of
water resources in this unique region.
This work was supported by the National Natural Science
Foundation of China (40830639) and the National Basic
Research Program of China ‘973 Program’
Copyright © 2015 John Wiley & Sons, Ltd.
(2010CB951101). The authors acknowledge the modelling groups, the Program for Climate Model Diagnosis
and Intercomparison (PCMDI) and the WCRP’s Working
Group on Coupled Modeling (WGCM) for their roles in
making available the WCRP CMIP3 multimodel data set.
The authors also thank the Institute of Soil Science,
Chinese Academy of Sciences (CAS), for providing soil
data and Cold and Arid Regions Environmental and
Engineering Research Institute, CAS, for providing land
use data.
Abbaspour KC, Faramarzi M, Ghasemi SS, Yang H. 2009. Assessing the
impact of climate change on water resources in Iran. Water Resources
Research 45: W10434. DOI: 10.1029/2008WR007615
Arnold JG, Srinivasan R, Muttiah RS, Allen PM. 2007. Continental scale
simulation of the hydrologic balance. Journal of the American Water
Resources Association 35(5): 1037–1051.
Ashraf Vaghefi S, Mousavi S, Abbaspour K, Srinivasan R, Yang H. 2014.
Analyses of the impact of climate change on water resources
components, drought and wheat yield in semiarid regions: Karkheh
River Basin in Iran. Hydrological Processes 28(4): 2018–2032.
Bastola S, Murphy C, Sweeney J. 2011. The role of hydrological
modelling uncertainties in climate change impact assessments of Irish
river catchments. Advances in Water Resources 34(5): 562–576.
Bennett KE, Werner AT, Schnorbus M. 2012. Uncertainties in hydrologic
and climate change impact analyses in headwater basins of British
Columbia. Journal of Climate 25(17): 5711–5730.
Bhat KS, Haran M, Terando A, Keller K. 2011. Climate projections using
Bayesian model averaging and space–time dependence. Journal of
Agricultural, Biological, and Environmental Statistics 16(4): 606–628.
Boé J, Terray L, Martin E, Habets F. 2009. Projected changes in
components of the hydrological cycle in French river basins during the
21st century. Water Resources Research 45:W08426.
Borah D, Yagow G, Saleh A, Barnes PL, Rosenthal W, Krug EC, Hauck
LM. 2006. Sediment and nutrient modeling for TMDL development
and implementation. Transactions of the ASAE 49(4): 967–986.
Bosshard T, Carambia M, Goergen K, Kotlarski S, Krahe P, Zappa M, Sch
€ar C. 2013. Quantifying uncertainty sources in an ensemble of
hydrological climate-impact projections. Water Resources Research 49:
Chang H, Jung IW. 2010. Spatial and temporal changes in runoff caused
by climate change in a complex large river basin in Oregon. Journal of
Hydrology 388(3): 186–207.
Chen L, Zheng H, David Chen Y, Liu C. 2008. Base-flow separation in
the source region of the Yellow River. Journal of Hydrologic
Engineering 13(7): 541–548.
Chen J, Brissette FP, Poulin A, Leconte R. 2011. Overall uncertainty
study of the hydrological impacts of climate change for a Canadian
watershed. Water Resources Research 47(12): W12509.
Cuo L, Zhang Y, Gao Y, Hao Z, Cairang L. 2013. The impacts of climate
change and land cover/use transition on the hydrology in the upper
Yellow River Basin, China. Journal of Hydrology 502: 37–52.
Dingman SL. 1994. Physical Hydrology. Englewood Cliffs, New Jersey:
Fontaine T, Cruickshank T, Arnold J, Hotchkiss R. 2002. Development of
a snowfall–snowmelt routine for mountainous terrain for the soil water
assessment tool (SWAT). Journal of Hydrology 262(1): 209–223.
Forbes KA, Kienzle SW, Coburn CA, Byrne JM, Rasmussen J. 2011.
Simulating the hydrological response to predicted climate change on a
watershed in southern Alberta, Canada. Climatic Change 105(3–4):
Fritsch J, Hilliker J, Ross J, Vislocky R. 2000. Model consensus. Weather
and Forecasting 15(5): 571–582.
Gao Q, Wan Y, Xu H, Li Y, Jiangcun W, Borjigidai A. 2010. Alpine
grassland degradation index and its response to recent climate
Hydrol. Process. 29, 4379–4397 (2015)
variability in Northern Tibet, China. Quaternary International 226(1–2):
Gao J, Zhang Y, Liu L, Wang Z. 2014. Climate change as the major driver
of alpine grasslands expansion and contraction: a case study in the Mt.
Qomolangma (Everest) National Nature Preserve, southern Tibetan
Plateau. Quaternary International 336(26): 108–116.
Giorgi F, Mearns LO. 2002. Calculation of average, uncertainty range, and
reliability of regional climate changes from AOGCM simulations via
the “reliability ensemble averaging” (REA) method. Journal of Climate
15(10): 1141–1158.
Horton P, Schaefli B, Mezghani A, Hingray B, Musy A. 2006. Assessment
of climate-change impacts on alpine discharge regimes with climate
model uncertainty. Hydrological Processes 20(10): 2091–2109.
Hu Y, Maskey S, Uhlenbrook S. 2013. Downscaling daily precipitation
over the Yellow River source region in China: a comparison of three
statistical downscaling methods. Theoretical and Applied Climatology
112(3–4): 447–460.
Immerzeel WW, van Beek LPH, Bierkens MFP. 2010. Climate change
will affect the Asian water towers. Science 328: 1382–1385.
IPCC. 2013. Working Group I Contribution to the IPCC Fifth Assessment
Report Climate Change 2013: The Physical Science Basis, Summary
for Policymakers (SPM).
uploads/WGI_AR5_SPM_brochure.pdf (Accessed May 17, 2014).
Jha M, Pan Z, Takle ES, Gu R. 2004. Impacts of climate change on
streamflow in the Upper Mississippi River Basin: a regional climate
model perspective. Journal of Geophysical Research 109: D09105.
Jin X, Sridhar V. 2012. Impacts of climate change on hydrology and water
resources in the Boise and Spokane River Basins. Journal of the
American Water Resources Association 48(2): 197–220.
Kay A, Davies H, Bell V, Jones R. 2009. Comparison of uncertainty
sources for climate change impacts: flood frequency in England.
Climatic Change 92(1–2): 41–63.
Li L, Hao Z, Wang J, Wang Z, Yu Z. 2008. Impact of future climate
change on runoff in the head region of the Yellow River. Journal of
Hydrologic Engineering 13(5): 347–354.
Li L, Shen H, Dai S, Xiao J, Shi X. 2012. Response of runoff to climate
change and its future tendency in the source region of Yellow River.
Journal of Geographical Sciences 22(3): 431–440.
Li F, Zhang Y, Xu Z, Teng J, Liu C, Liu W, Mpelasoka F. 2013. The
impact of climate change on runoff in the southeastern Tibetan Plateau.
Journal of Hydrology 505: 188–201.
Majone B, Bovolo C, Bellin A, Blenkinsop S, Fowler H. 2012. Modeling
the impacts of future climate change on water resources for the Gállego
river basin (Spain). Water Resources Research 48(1): W01512.
Maurer E, Adam J, Wood A. 2009. Climate model based consensus on the
hydrologic impacts of climate change to the Rio Lempa basin of Central
America. Hydrology and Earth System Sciences 13: 183–194.
Min SK, Hense A. 2006. A Bayesian approach to climate model
evaluation and multi-model averaging with an application to global
mean surface temperatures from IPCC AR4 coupled climate models.
Geophysical Research Letters 33(8): L08708.
Min SK, Park E, Kwon WT. 2004. Future projections of East Asian
climate change from multi-AOGCM ensembles of IPCC SRES scenario
simulations. Journal of the Meteorological Society of Japan 82(4):
Min SK, Simonis D, Hense A. 2007. Probabilistic climate change
predictions applying Bayesian model averaging. Philosophical Transactions of the Royal Society A: Mathematical, Physical and
Engineering Sciences 365(1857): 2103–2116.
Moradkhani H, Baird RG, Wherry SA. 2010. Assessment of climate
change impact on floodplain and hydrologic ecotones. Journal of
Hydrology 395(3): 264–278.
Moriasi D, Arnold J, Van Liew M, Bingner R, Harmel R, Veith T. 2007.
Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Transactions of the ASABE 50(3): 885–900.
Neitsch S, Arnold J, Kiniry J, Williams J. 2005. Soil and Water Assessment
Tool Theoretical Documentation Version 2005.
media/1292/SWAT2005theory.pdf (Accessed Jan 15, 2014).
Nohara D, Kitoh A, Hosaka M, Oki T. 2006. Impact of climate change on
river discharge projected by multimodel ensemble. Journal of
Hydrometeorology 7: 1076–1089.
Copyright © 2015 John Wiley & Sons, Ltd.
Palmer T et al. 2004. Development of a European multimodel ensemble
system for seasonal-to-interannual prediction (DEMETER). Bulletin of
the American Meteorological Society 85(6): 853–872.
Palmer T et al. 2005. Representing model uncertainty in weather and
climate prediction. Annual Review of Earth and Planetary Sciences 33:
Paton F, Maier H, Dandy G. 2013. Relative magnitudes of sources of
uncertainty in assessing climate change impacts on water supply
security for the southern Adelaide water supply system. Water
Resources Research 49: 1643–1667.
Phillips TJ, Gleckler PG. 2006. Evaluation of continental precipitation in
20th century climate simulations: the utility of multimodel statistics.
Water Resources Research 42(3): W03202.
Prudhomme C, Reynard N, Crooks S. 2002. Downscaling of global
climate models for flood frequency analysis: where are we now?
Hydrological Processes 16(6): 1137–1150.
Qian J, Wang G, Ding Y, Liu S. 2006. The land ecological evolutional
patterns in the source areas of the Yangtze and Yellow Rivers in the
past 15 years, China. Environmental Monitoring and Assessment
116(1–3): 137–156.
Raftery AE, Gneiting T, Balabdaoui F, Polakowski M. 2005. Using
Bayesian model averaging to calibrate forecast ensembles. Monthly
Weather Review 133(5): 1155–1174.
Setegn SG, Rayner D, Melesse AM, Dargahi B, Srinivasan R. 2011.
Impact of climate change on the hydroclimatology of Lake Tana Basin,
Ethiopia. Water Resources Research 47: W04511.
Sharpley A, Williams J. 1990. EPIC-Erosion Productivity Impact
Calculator, 1. model documentation. U.S. Department of Agriculture,
Agriculture Research Service. Technical Bulletin. No.1768.
Teutschbein C, Wetterhall F, Seibert J. 2011. Evaluation of different
downscaling techniques for hydrological climate-change impact studies
at the catchment scale. Climate Dynamics 37(9–10): 2087–2105.
Wang G, Li Y, Chen L, Wu Q, Wang Y. 2006. Impacts of permafrost
changes on alpine ecosystem in Qinghai-Tibetan Plateau. Science in
China Series D: Earth Sciences 49(11): 1156–1169.
Wang G, Hu H, Li T. 2009. The influence of freeze-thaw cycles of active
soil layer on surface runoff in a permafrost watershed. Journal of
Hydrology 375(3–4): 438–449.
Wang G, Bai W, Li N, Hu H. 2011a. Climate changes and its impact on
tundra ecosystem in Qinghai-Tibet Plateau, China. Climate Change
206(3): 463–482.
Wang J, Wang E, Li LD. 2011b. Modelling the impacts of climate change
on wheat yield and field water balance over the Murray–Darling Basin
in Australia. Theoretical and Applied Climatology 104(3–4): 285–300.
Wang X, Yang T, Shao Q, Acharya K, Wang W, Yu Z. 2012. Statistical
downscaling of extremes of precipitation and temperature and construction of their future scenarios in an elevated and cold zone. Stochastic
Environmental Research and Risk Assessment 26(3): 405–418.
Wilby RL, Harris I. 2006. A framework for assessing uncertainties in
climate change impacts: low-flow scenarios for the River Thames, UK.
Water Resources Research 42(2): W02419.
Wood AW, Maurer EP, Kumar A, Lettenmaier DP. 2002. Long-range
experimental hydrologic forecasting for the eastern United States.
Journal of Geographical Research 107(D20): 4429.
Wood AW, Leung LR, Sridhar V, Lettenmaier DP. 2004. Hydrologic
implications of dynamical and statistical approaches to downscaling
climate model outputs. Climatic Change 62(1): 189–216.
Wu Q, Li X, Li W. 2000. The prediction of permafrost change along the
Qinghai-Tibet Highway, China. Permafrost and Periglacial Processes
11(4): 371–376.
Xie H, Ye J, Liu X, Chongyi E. 2010. Warming and drying trends on the
Tibetan Plateau (1971–2005). Theoretical and Applied Climatology
101(3–4): 241–253.
Xu Z, Zhao F, Li J. 2009. Response of streamflow to climate change in the
headwater catchment of the Yellow River basin. Quaternary International 208(1): 62–75.
Yang T, Wang X, Zhao C, Chen X, Yu Z, Shao Q, Xu C, Xia J, Wang W.
2011. Changes of climate extremes in a typical arid zone: observations
and multimodel ensemble projections. Journal of Geophysical Research
116(D19): D19106.
Yang Z, Gao J, Zhao L, Xu X, Ouyang H. 2013. Linking thaw depth with
soil moisture and plant community composition: effects of permafrost
Hydrol. Process. 29, 4379–4397 (2015)
degradation on alpine ecosystems on the Qinghai-Tibet Plateau. Plant
and Soil 367(1–2): 687–700.
Yi X, Li G, Yin Y. 2012. Temperature variation and abrupt change
analysis in the Three-River Headwaters Region during 1961–2010.
Journal of Geographical Sciences 22(3): 451–469.
Yin F, Deng X, Jin Q, Yuan Y, Zhao C. 2014. The impacts of climate
change and human activities on grassland productivity in Qinghai
Province, China. Frontiers of Earth Science 8(1): 93–103.
Zhang X, Srinivasan R, Debele B, Hao F. 2008. Runoff simulation of the
headwaters of the Yellow River using the SWAT model with three
snowmelt algorithms. Journal of the American Water Resources
Association 44(1): 48–61.
Zhang Y, Wang G, Wang Y. 2010. Response of biomass spatial pattern of
alpine vegetation to climate change in permafrost region of the QinghaiTibet Plateau, China. Journal of Mountain Science 7(4): 301–314.
Zhang S, Hua D, Meng X, Zhang Y. 2011. Climate change and its driving
effect on the runoff in the “Three-River Headwaters” region. Journal of
Geographical Sciences 21(6): 963–978.
Copyright © 2015 John Wiley & Sons, Ltd.
Zhang Y, Zhang S, Zhai X, Xia J. 2012. Runoff variation and its response
to climate change in the Three Rivers Source Region. Journal of
Geographical Sciences 22(5): 781–794.
Zhang L, Su F, Yang D, Hao Z, Tong K. 2013. Discharge regime and
simulation for the upstream of major rivers over Tibetan Plateau.
Journal of Geophysical Research, [Atmospheres] 118(15): 8500–8518.
Zhang P, Liu R, Bao Y, Wang J, Yu W, Shen Z. 2014. Uncertainty of
SWAT model at different DEM resolutions in a large mountainous
watershed. Water Research 53: 132–144.
Zhao F, Xu Z, Huang J, Li J. 2008. Monotonic trend and abrupt changes
for major climate variables in the headwater catchment of the Yellow
River basin. Hydrological Processes 22(23): 4587–4599.
Zhou D, Huang R. 2012. Response of water budget to recent climatic
changes in the source region of the Yellow River. Chinese Science
Bulletin 57(17): 2155–2162.
Zhou H, Zhao X, Tang Y, Gu S, Zhou L. 2005. Alpine grassland degradation
and its control in the source region of the Yangtze and Yellow Rivers,
China. Japanese Society of Grassland Science 51(3): 191–203.
Hydrol. Process. 29, 4379–4397 (2015)
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF